From .Net Developer to AI Engineer

You already have most of the skills an AI engineer needs. This book teaches you the rest, in C#.

Job titles that used to say "Senior .NET Developer" increasingly say "AI Engineer." But an AI engineer is not a data scientist. AI engineers don't train models; they take existing models and turn them into dependable software. That means API design, dependency injection, data access, security boundaries, testing, observability and deployment: work you already do every day.

From .NET Developer to AI Engineer is a practical, opinionated guide for experienced .NET developers who need to ship AI features that hold up in production, not demos that fall over the first time a real customer uses them. You don't need a background in machine learning, heavy maths or Python. Every line of code is C# on .NET 10.

Build one real system, chapter by chapter
You'll build Northwind Assist, a customer-support assistant that grows from a simple chat endpoint into a grounded, tool-using, multi-agent system with security, evaluation and telemetry. Every chapter comes with runnable companion code on GitHub.

What you'll learn

• How LLMs really work (tokens, context windows, embeddings, sampling) and how to choose a model on capability, cost and latency
• Prompt engineering as an engineering discipline, with reusable templates, versioning and testing
• Microsoft.Extensions.AI: IChatClient, streaming, middleware and switching between Azure OpenAI, OpenAI and Ollama without rewriting your app
• Structured outputs mapped to strongly typed C# models, plus tool calling that safely exposes your APIs and business services
• Running open models locally with Ollama
• Vector databases with Microsoft.Extensions.VectorData, and retrieval-augmented generation (RAG) end to end: chunking, filtering, citations and reranking
• Microsoft Agent Framework, the successor to Semantic Kernel and AutoGen: agents, sessions, memory, workflows and migration
• Model Context Protocol (MCP): building and consuming MCP servers in .NET
• Multi-agent orchestration and handoffs, while keeping agents bounded and predictable
• AI security mapped to the OWASP Top 10 for LLMs, including prompt injection and excessive agency
• Quality evaluation with Microsoft.Extensions.AI.Evaluation, run in CI/CD
• Observability with OpenTelemetry: tracing, token usage, cost dashboards and alerts
• Production architecture: resilience, rate limits, caching, fallbacks, deployment, cost and governance

Who this book is for

• Experienced .NET developers ready to move into AI engineering
• Senior engineers and tech leads who have been asked to "add AI" and need it to work in production
• Solution architects making decisions about providers, data flow, cost and quality
• Developers coming from Semantic Kernel or AutoGen who want to understand today's Microsoft AI stack

The book closes with a career chapter that maps your existing cloud, security, data and architecture experience to AI engineering roles, and gives you a portfolio plan to prove it.

About the author
Najaf A. Shaikh is a software architect with more than twenty years of experience building enterprise systems on .NET, including over a decade leading the architecture of large-scale wealth-management platforms across the UK, the US and Australia.